0354 98361100041N

Naeimi, H., et al.
: A Parametric Design of Compact Exhaust Manifold Junction in

THERMAL SCIENCE, Year 2011, Vol. 15, No. 4, pp. 1023-1033 1023

A PARAMETRIC DESIGN OF COMPACT EXHAUST MANIFOLD
JUNCTION IN HEAVY DUTY DIESEL ENGINE USING
COMPUTATIONAL FLUID DYNAMICS CODES
by
Hessamedin NAEIMI
a*
, Davood DOMIRY GANJI
a

Mofid GORJI
a
, Ghasem JAVADIRAD
b
, and Mojtaba KESHAVARZ
b

a
Faculty of Mechanical Engineering, Babol University of Technology, Babol, Iran
b
Iran heavy diesel MFG. Co. (DESA), Amol, Iran
Original scientific paper
UDC: 621.43.041.6:519.872
DOI: 10.2298/TSCI100417041N
Nowadays, computational fluid dynamics codes are prevalently used to simulate
the gas dynamics in many fluid piping systems such as steam and gas turbines,
inlet and exhaust in internal combustion engines. In this paper, a computational
fluid dynamics software is used to obtain the total energy losses in adiabatic
compressible flow at compact exhaust manifold junction. A steady-state one-
dimensional adiabatic compressible flow with friction model has been applied to
subtract the straight pipe friction losses from the total energy losses. The total
pressure loss coefficient has been related to the extrapolated Mach number in the
common branch and to the mass flow rate ratio between branches at different
flow configurations, in both combining and dividing flows. The study indicate
that the numerical results were generally in good agreement with those of expe-
rimental data from the literature and will be applied as a boundary condition in
one-dimensional global simulation models of fluid systems in which these com-
ponents are present.
Key words: Diesel engines, exhaust manifold junction, pressure loss coefficient,
numerical simulation, experimental data
Introduction
The appropriate selection of turbocharging system type and the reasonable design of
exhaust manifold configurations in heavy-duty Diesel engines is very significant since the
performance of a four-stroke turbocharged Diesel engine is greatly affected by the gas flow in
the exhaust manifold [13]. At the present time, several different turbocharging systems are
usually adopted: the constant pressure turbocharging system, the pulse turbocharging system,
the pulse converter (PC) turbocharging system, and the compact exhaust manifold or modular
pulse converter (MPC) turbocharging system, etc. In the constant pressure turbocharging
system, the exhaust ports of all cylinders are connected to a single manifold to damping
unsteady gas flow from cylinders. Hence, the pressure in the turbine inlet is almost steady.

Corresponding author: e-mail: hessam.naeimi@gmail.com
Naeimi, H., et al.: A Parametric Design of Compact Exhaust Manifold Junction in
1024 THERMAL SCIENCE, Year 2011, Vol. 15, No. 4, pp. 1025-1035

This allows the turbine to operate at optimum efficiency at specified engine conditions. This
matter is a major advantage of this type of turbocharging. However, the significant
disadvantages of the constant pressure turbocharging are poor turbocharging acceleration and
performance at low speed and load. In the case of the pulse turbocharging system, the exhaust
gases coming from two or three cylinders, which have minimum interference in scavenging
process base on the firing order, are discharged into a common branch exhaust pipe. It aims to
make maximum use of the energy available in the high pressure and temperature exhaust
gases. This turbocharging system has good turbocharger acceleration and performance at low
speed and part load.
However, its application is limited by the number of the exhaust manifold and the
design of the exhaust manifold with large numbers of cylinders is complex. In addition the
turbine efficiency with one or two cylinder per turbine entry or at very high rating is poor [4].
Pulse converter turbocharging system has been developed to overcome the disadvantage of
the pulse turbocharger. In this system, many cylinders are connected to a single low volume
manifold. So that the pressure variation at the turbine entry is reduced and its result the
turbine efficiency is improved. However, the PC turbocharging system has poor performance
at very low speed/load and is only suitable for engines with certain numbers of cylinders (e. g.
four, eight, sixteen, etc.) [5]. Nowadays, the
MPC turbocharging system has become
popular because the structure of its exhaust
manifolds is simple. In this kind of turbo-
charging system, as it is shown in fig. 1, all
the exhaust ports are connected to a common
exhaust duct by a distinct pulse converter.
This system intends to preserve the pulsing
energy of exhaust gases coming from
cylinders and transmit it to the turbine inlet
while reduces the backflow toward cylinders
during scavenging as possible [6].
Nowadays, the 1-D models are used in the global simulation of steady and transient
compressible flow in pipe systems, such as BOOST (AVL) [7], and GT POWER [8]. These
models are utilized as well as to analysis of internal combustion engine performance at
various operating conditions. However, the fluid flow through the compact exhaust manifold
junctions, as its geometry, is 3-D and highly turbulent. Since the total pressure loss
coefficients must be obtained separately and added to these models as special boundary.
Unfortunately there is not enough pressure loss data in literature especially for compressible
flow in compact exhaust manifold. The largest source of the experimental result for pressure
loss in T junction have been perform by Miller [9]. Basset et al. [10] compared different
modelling techniques. A multi-dimensional computer program was developed by Chiatti et al.
[11]. These cods were used to simulate the flow within different components of the exhaust
system in internal combustion engine (ICE). Commercial codes, such as Star-CD, Fluent, or
Fire-AVL were used by Shaw et al. [12], and Gan et al. [13]. Abou-Haidar et al. [14] and
Pearson et al. [15] utilized 1-D and 2-D models to simulate the wave propagation
phenomenon. Most of the works have been focused on designing manifolds of ICE. In the
present work, a commercial computational fluid dynamics (CFD) package, FLUENT [16] is
used to obtain the total energy losses in adiabatic compressible flow at compact exhaust

Figure 1. Modular pulse converter junction of
Chan [17]

manifold junction. The numerical results were generally in good agreement with the steady
flow measurements of Chan [17].
Mathematical model
The fluid flow studied is governed by three-dimensional compressible adiabatic
steady-state form of the Reynolds-averaged Navier-Stokes (RANS) conservation equations
and the additional equations describing the transport of other scalar properties. They may be
written in Cartesian tensor notation as:

( ) 0
i
i
u
x

(1)

k
k
eff
2
( ) ( )
3
j
i
i j ij i j
i i j j i j
i j
u
u u p
u u u u
x x x x x x x
p
x x

(2)

0 eff eff
( ) ( )
i i i
i j j
u h u u
x x x
(3)

Different turbulence models based on RANS equations is available for studying the
Reynolds stresses and the turbulent diffusivity terms. In this work, the two equations turbulent
model k-eRNG will be used and different turbulent models were compared. The random
number generation (RNG) model has an additional term in its equation that significantly
improves the accuracy for rapidly strained flows. The effect of swirl on turbulence is included
in the RNG model, enhancing accuracy for swirling flows. The turbulence kinetic energy, k,
and its rate of dissipation, e, are obtained from the following transport equations:

t
k
( )
div( U) div 2
t ij ij
k
k E E
t
(4)

2
t
1 t 2
( )
div( ) div grad 2
ij ij
U C E E C
t k k
(5)

The coefficients of eqs. (4) and (5) are shown
in tab. 1.
Definition of junction pressure
loss coefficient
The studying here for calculation of the pressure loss at compact exhaust manifold
will be limited by duct configuration and flow direction. Actually, there are six individual
flow types for a generalized three-pipe junction (fig. 2) and twelve loss coefficients associated

2
t
0.0845
k
C C
(6)
Table 1. Coefficients of k-e RNG model
C
1e
C
2e
s
k
s
e

1.42 1.68 1.0 1.3


with it. In this study (as shown in fig. 2), we
have considered two flow types, one from
combining flow type and another dividing
flow.
The total pressure loss coefficient K
ij
, given
by eq. (7), is defined as the ratio of the total
pressure drop in the direction of positive mass
flow to the dynamic-pressure (P
d
) in the
common branch that witch the total flow is
passing [9]:

where

t
if
if
i i j
j i j
u Q Q
u
u Q Q
(8)

Notice that the dynamic-pressure (P
d
) in the leg carrying the total flow for
compressible flow is obtained from:

* *
d 0t st
P P P (9)

So, the four loss coefficients for the combining and dividing flow as shown in fig. 2
are calculated as:

combining flow

2 2
1 s1 3 s3
13
13
*
3 03 3
1 1
2 2
u P u P
P
K
P P P

(10)

2 2
2 s2 3 s3
23
23
*
3 03 3
1 1
2 2
u P u P
P
K
P P P

(11)
dividing flow

2 2
3 s3 1 s1
31
31
*
3 03 3
1 1
2 2
u P u P
P
K
P P P

(12)

2 2
3 s3 2 s2
32
32
*
3 03 3
1 1
2 2
u P u P
P
K
P P P

(13)

When the Mach number of air flow is less than 0.2, the air passing of compact
exhaust manifold can be considered incompressible. The loss coefficient for incompressible

Figure 2. Schematic of modular pulse
converter junction

2 2
si s
2
d
t
1 1
2 2
1
2
i j i
ij
ij
u P u P
P
K
P
u
(7)

flow is independent of common branch much number and just changes with mass flow ratio.
However, the value of loss coefficient for compressible flow does vary with common branch
much number, too.
Computational domain
The flow domain as shown in fig. 3 has a three-pipe intersection. An unstructured,
non-uniform mesh, a detail of the grid in the intersection of the branches region, the mesh
structure on the symmetry plane and a detail of the mesh at the cross-section are illustrated.
Due to the symmetry of modular exhaust junction, in all flow configurations the
computational domain can be considered one half of the actual volume. In order to make sure
that the accuracy of results was independent of computations grid, calculations were carried
out with several different grid resolutions as well as modifying the distance of the wall-
adjacent cells to the wall. For all mass flow rates simulated, the y
+
value is maintained within
the recommended range, 30 < y
+
< 500. The number of cells of the mesh finally used was
186,354.

Figure 3. 3-D computational domain, mesh structure in the plane of symmetry and in the
cross-sectional of a branch at compact manifold junction simulated

In all of the studied cases here, the much number of the outlet exhaust manifold is
assumed constant and is 0.25. Outlet pressure and temperature was fixed of 300 Pa and 300 K,
respectively. So, based on the target mass flow rate ratio (q), desired mass flow rate for each
branch has been studied. Since the air flow was supposed compressible, a mass flow rate and
static pressure at the inlet and outlet boundary condition has been chosen, respectively. Also,
for simulating the turbulent flow, the turbulence intensity and the hydraulic diameter were set
to 3.5% and 50 mm, respectively, as turbulence parameters in both of the combining and
dividing flow. Flow is adiabatic and non-slip condition with wall roughness height is used for
wall condition.
Evaluating the pipe wall friction factor
In estimating pressure loss due to compact exhaust manifold, the pressure
measurements location is great importance. The pressure should be measured in fully
developed flow region. So that the one dimensional Fanno flow model could be used. In other

words, if the pressure measuring location was
not selected correctly the result would not be
reliable. Figure 4 demonstrates the total pres-
sure changes in symmetry surface of different
compact exhaust manifolds having different
branch lengths.
As can be seen from fig. 4, the total pressure
of the flow in manifolds having length equal or
greater then 18D remains constant. Thus, the
flow condition for compact exhaust manifold
having aforesaid branch length will be fully
developed. The pipe friction factor is defined in
eq. (14) as:

w
2
1
2
f
u

(14)

It is common practice, in wave-action simulation, to use a constant value of f

in the
region of 0.004-0.01. In fact, the curve on the Moody diagram for a smooth pipe (D 2.5 mm)
gives value in the range of 0.0035 to 0.008 for Reynolds numbers in the range of 110
4
to
510
5
. For Reynolds numbers in the range of 510
3
Re 10
8

and relative roughness values
in the range of 10
6
D/D 10
2
, Swamee et al. [18] propose the following relationship:

2
10
0.25
5.74
log
3.7 Re
f
D

(15)

Equation (15) can be applied to give either a value for at every mesh point and time
step of the calculation or to give an average value for each point section comprising manifold
[19].
Using 1-D and adiabatic flow conditions between two points in Fanno flow, Mach
number can be calculated from eq. (16). Then all thermo-fluid properties extrapolated can be
obtained and the total pressure loss coefficient computed from eq. (7):

2
2
2
2 1 2
2 2 2
2
2 1 1
1
1
1
1 1 1 1 1
2
ln ln
1
2 2
1
2
M
x x M
f
D
M M M
M
(16)
Results and discussion
Different turbulent models implemented in fluent have been compared. The k-e
standard and RNG with standard wall functions, k-w

standard and Spalart-Allmaras
turbulence models have been used in preliminary simulations at different flow configurations.
In fig. 5, the total pressure loss coefficients obtained with each turbulence model are
represented against a mass flow rate between branches and compare with experimental data
by Chan [17].

Figure 4. Total pressure curve in symmetry
surface of common branch

It can be observed that in the both
flow types, loss coefficient predicted by
k-e RNG turbulence model is in a good
agreement with reference data. As a
result, this turbulence model will be used
in all simulations.
Combining flow
The compact exhaust manifold with
combining flow type has been simulated
with different mass flow rate ratio, q = 0.1
to 0.9, for combining flow type. Figure 6
shows the simulation results on the
velocity contour on the symmetry plane
in the compact manifold near the junc-
tion for three flow ratios.

Figure 5. Comparison between different turbulence
models (combining flow)

Figure 6. Predicted combining flows
for the centre plane of the compact
manifold


Mainly, the multi-dimensional undesirable phenomena, such as flow separation,
turbulent mixing, eddies production etc. caused to decrease the total energy of fluid flow. By
increasing the mass flow rate in branch 1, the flow velocity increases in throat area and then
the effect of flow separation and vortex formation due to the changing cross section in flow
upstream increases. So, the pressure loss coefficient is increased. Most published
experimental data for pressure loss coefficients are for T-junctions with a straight branch. The
only data for compact exhaust manifold comparable with the present investigation is that
reported by Chan [17]. A comparison between the predicted and reported pressure loss
coefficients is given in tab. 2. The predicted pressure loss coefficient is in good agreement
with the reported value.

Table 2. Comparison of junction pressure loss coefficients for combining flow

Throat area (At) is one of the important
parameters in compact manifold design.
Compact exhaust manifold works as fluid
diode which allows exhaust gas to move from
cylinder to the turbine and limits the flow
returning from manifold to the cylinder. This
will be done by decreasing the throat area.
With this decrement in throat area, the
velocity will be increased and the pressure will
be decreased. So, it will avoid back pressure in
exhaust manifold hence flow returning to the
cylinder will be limited. Hence, in this study
the effect of the area ratio of At/Ap on the
pressure loss coefficient has been investigated.
Figures 7 and 8 show the variation of the loss
coefficients K
13
and K
23
with mass flow rate
for three different At/Ap of 0.3, 0.35, and 0.4.
As shown in figs. 7 and 8, the pressure loss coefficient in each branch will be increased with
flow rate. In lower flow rates, the ratio of At/Ap has not significant effect on the pressure loss
coefficient. However, in the higher flow ratio of the branch 1 to branch 3, K
13
will be
increased with At/Ap decrease. The situation happens in reverse form for K
23
. It can be con-
q
K
13
K
23

Predicted Exp. [17] Predicted Exp. [17]
0.1 0.7 n. a. 3.44 n. a.
0.2 0.21 0.29 2.16 1.97
0.3 1.2 n. a. 1.7 n. a.
0.4 2.01 2.31 1.11 1.28
0.5 3.19 n. a. 0.74 n. a.
0.6 5.2 5.64 0.35 0.23
0.7 7.46 n. a. 0.05 n. a.
0.8 10.4 11.88 0.32 0.48
0.9 13.57 n. a. 0.52 n. a.

Figure 7. Effect of At/Ap on pressure loss
coefficient K
13


cluded from above that in a specific flow rate,
magnitude of K
23
and K
13
will be decreased
with At/Ap ratio increase.
Dividing flow
Figure 9 shows the predicted dividing
flow-patterns near the intersection for three
different flow ratios. In low mass flow rate
ratio (q = 0.1) eddies are seen in side branch
(branch 1 ). By addition of flow ratio (q =
= 0.5), the location of recirculation region in
the branch changes and velocity increases in
throat. By further increases of velocity, we
accost chock fig. 9(c). The hoped-for full
range up to q = 0.9 could not be obtained
despite adequate pressure being available
upstream. In dividing flow, q ranged from 0.1 up to a maximum of about 0.6, which was
effectively the choking limit. The predicted and reported pressure loss coefficients for
dividing flows for flow ratios up to 0.6 is shown in tab. 3.

Figure 8. Effect of At/Ap on pressure loss
coefficient K
23

Figure 9. Predicted dividing flows for the
centre plane of the compact manifold

Table 3. Comparison of junction pressure loss coefficients for combining flow

As seen from tab. 3, the simulation results are in good agreement with experimental
data. For mass flow rate ratio more than 0.6, a shock wave is developed when the supersonic
flow passes through the side branch 1 . So, the pressure loss coefficient can not be obtained
for larger mass flow rate.
Conclusions
In this paper a computational fluid flow model, k- RNG, has been used to obtain
the pressure loss coefficient in compact exhaust manifold junction. Also, to subtract the
frictional losses from the total energy losses, the 1-D Fanno flow model has been used. A
comparison between the predicted and experimental data shows this model and methodology
is in generally good for the estimate of the pressure losses in both combining and dividing
flow. But the maximum mass flow rate of the divided flow was 0.6, because a limit caused by
chocking.
Finally, these coefficients can be used as boundary conditioning in 1-D software
such as GT-Power to simulate the complete engine cycle. As a suggestion for future work, we
can optimize the design parameters of the compact manifold.
Nomenclature
A pipe cross-sectional area, [m
2
]
At trot area, [m
2
]
Ap area at P cross-sectional, [m
2
]
D internal diameter, [m]
E
ij
rate of deformation, [s
1
]
f friction factor, []
h
0
stagnation enthalpy, [Jkg
1
]
K total pressure loss coefficient, []
k turbulent kinetic energy, [m
2
s
1
]
M total pressure loss coefficient, []
P
s
static pressure, [Pa]
P
0
stagnation pressure, [Pa]
Q mass flow rate, [kgs
1
]
q mass flow rate ratio, []
Re Reynolds number, []
U average gas velocity, [ms
1
]
u
i
time-averaged gas velocity, [ms
1
]
y
+
sublayer scaled distance y
+
= u
y/, []

Greek symbols
ratio of specific heats = c
p
/c
v
, []
non-dimensional roughness, [m]
turbulent dissipation rate, [m
2
s
3
]
absolute viscosity, [Pas]
t
turbulent viscosity, [Pas]
gas density, [kgm
3
]
eff
apparent stress tensor, [Pa]
v
Rayleigh dissipation function, [Pas
1
]
ij
Kronecker delta, []
junction lateral branch angle
Subscripts
i,j Branch leg index
* Extrapolated properties to the junction
1, 2 Inlet (combining flow) or outlet
(dividing flow) branches
3 Common branch
q
K
31
K
32

Predicted
Exp. [17]
Predicted
Exp. [17]
0.1 0.46 n. a. 0.386 n. a.
0.2 1.11 1.53 0.29 0.316
0.3 1.8 n. a. 0.28 n. a.
0.4 3.5 3.92 0.3 0.285
0.5 6.0 n. a. 0.32 n. a.
0.6 10.6 9.57 0.35 0.309

References
[1] Kesgin, U., Effect of Turbocharging System on the Performance of a Natural Gas Engine, Energy
Conversion and Management, 46 (2005), 1, pp. 11-32
[2] Galindo, J., et al., Design of an Exhaust Manifold to Improve Transient Performance of a High-Speed
Turbocharged Diesel Engine, Experimental Thermal and Fluid Science, 28 (2004), 8, pp. 863-875
[3] Kesgin, U., Study on the Design of Inlet and Exhaust System of a Stationary Internal Combustion
Engine, Energy Conversion and Management, 46 (2005), 13-14, pp. 2258-2287
[4] Hu, X., Lawless, P. B., Predictions of On-Engine Efficiency for the Radial Turbine of a Pulse
Turbocharged Engine. SAE paper 2001-01-1238, 2001
[5] Bassett, M. D., Pearson, R. J., Winterbone, D. E., Steady-Flow Loss-Coefficient Estimation for Exhaust
Manifold Pulse Converter Type Junctions, SAE paper 1999-01-0213, 1999
[6] Yang, S. Y., Deng, K. Y., Cui, Y., Simulation and Experimental Research on a Mixed Pulse Converter
Turbocharging System, Journal of Automobile Engineering, 221 (2007), 2, pp. 215-223
[7] ***, BOOST, User manual for AVL BOOST, AVL List GmbH, Department for Applied Thermo-
dynamics, 1996, http://www.avl.com
[8] ***, GT Power Users manual. GT Suite Version 5.1. Gamma Technologies Inc., USA, 2000
[9] Miller, D. S., Internal Flow Systems, 2
nd
ed., Craneld: The British Hydromechanics Research
Association, Cranfield, UK, 1990
[10] Basset, M. D., Winterbone, D. E., Pearson, R. J., Modelling Engines with Pulse Converted Exhaust
Manifolds using One-Dimensional Techniques, SAE paper 2000-01-0290, 2000
[11] Chiatti, G., Chiavola, O., Multicode Prediction of the Influence of the Exhaust System on the
Performance of a Turbocharged Engine, J. Eng. Gas Turbines Power, 124 (2002), 3, pp. 695-701
[12] Shaw, C. T., et al., Modelling the Effect of Plenum-Runner Interface Geometry on the Flow through an
Inlet System, SAE paper 2000-01-0569, 2000
[13] Gan, G., Riffat, S. B., Numerical Determination of Energy Losses at Duct Junctions, Appl. Energy 67
(2000), 3, pp. 331-340
[14] Abou-Haidar, N. I., Dixon, S. L., Pressure Losses in Combining Subsonic Flows through Branched
Ducts, Trans. ASME, J.Turbomach, 114 (1992), 1, pp. 264-270
[15] Pearson, R. J., et al., Multidimensional Wave Propagation in Pipe Junctions, SAE paper 1999-01-1186,
1999
[16] ***, FLUENT User's Guide, Fluent Inc., New York, USA, 1995
[17] Chan, C. L., An Investigation of the Performance of a Diesel Engine Fitted with Compact Manifold, M.
Sc. thesis, University of Manchester, Institute of Science and Technology, Manchester, UK, 1985
[18] Swamee, P. K., Jain, A. K., Explicit Equations for Pipe-Flow Problems, J. Hydraulic Div. Proc. ASCE,
102 (1976), 5, pp. 657-664
[19] Winterbone, D. E., Pearson, R. J., Theory of Engine Manifold Design. Wave Action Methods for IC
Engines, Professional Engineering Publishing, London, 2000

Paper submitted: April 17, 2011
Paper revised: May 3, 2011
Paper accepted: May 6, 2011

0354 98361100041N

Hochgeladen von

Dokumentinformationen

Originalbeschreibung:

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

0354 98361100041N

Hochgeladen von

Copyright:

Verfügbare Formate

Naeimi, H., et al.

: A Parametric Design of Compact Exhaust Manifold Junction in

Das könnte Ihnen auch gefallen